https://gcc.gnu.org/bugzilla/show_bug.cgi?id=71646
--- Comment #1 from Andrew Pinski <pinskia at gcc dot gnu.org> --- sm_30 is definitely the min target for offloading for GCC to nvptx . What I don't know if `nvidia geforce gt 430` support is still existant in cuda. Maybe someone who knows the offloading support for Nvidia GPUs should comment here really.